[ih] ARPANET pioneer Jack Haverty says the internet was never finished

[Karl Auerbach]

On 3/1/22 7:27 PM, the keyboard of geoff goodfellow via Internet-history 
wrote:

> curious if anyone has a copy or memory of what
> the Long List of Things That Had To Be Figured Out Someday?

My list begins with an association protocol layer - what ISO/OSI called 
the "session" layer, but nothing as incomprehensibly overburdened and 
extensive as what they had.

On the net we have a lot of "association" context that can transcend the 
lifetime of a single transport connection.

That sort of thing would have been useful to avoid the triangular 
routing used in mobile IP.

And it would have obviated a lot of cookie usage on the web.

Same for crypto context in things like DTLS.

Basically an association layer would let the end applications 
established named markers during their conversation.  Then when a 
transport broke and was re-established the two end-points would say 
"where were we when we last spoke?"  They would use the association 
protocol to find the last agreed-upon name.  It would be up to the 
applications to remember what they each did after that name and how to 
deal with it - sort of like a journaled file system.

Such a protocol would not require any storage in the protocol stack - 
that would be in the applications who need to remember what they did 
after the last agreed-upon named checkpoint.

	--karl--